Overview of Interval
The Interval HUE Platform: Harvest, U-AI, and Exchange.
The Enterprise-Grade Data Intelligence Platform
Transforming Enterprise Data into a Strategic Asset
Enterprises today are rich in data but poor in insights. Critical information is often siloed across disparate systems, difficult to access, and risky to share, preventing its use as a strategic asset. The Interval platform provides a vertically integrated data intelligence solution that runs alongside your infrastructure, transforming raw, fragmented data into trusted, high-impact intelligence while ensuring you maintain complete control and sovereignty.
Architectural Pillars
The entire platform operates on a Private Kubernetes Cluster within your existing cloud or on-premises environment, guaranteeing that your data never leaves your control. Data transit is secured by a multi-layered, zero-trust model that combines IPsec VPN tunnels for network transport security with end-to-end TLS Encryption. A granular Network Gateway acts as a final inspection and authorization point, ensuring every byte of data is protected at every stage of its journey.
Our architecture refines data through a proven progression—Standardized, Interval Data Standard, and Economic Performance Metrics—built on the robust, ACID-compliant foundation of Apache Iceberg. Unlike traditional ETL pipelines, Interval interleaves a powerful AI Intelligence Layer at each stage to automate and elevate data quality.
Standardized Layer: Raw data is ingested and immediately organized with AI-driven normalization and anonymization, establishing a clean, structured foundation for all future processing.
Interval Data Standard Layer: Data is enriched with AI-powered identification, quality enhancements, contextual validation, and roll up tables – transforming raw information into reliable, queryable assets.
Governance and Trust
A centralized Metadata Lake, powered by OpenMetadata, provides a unified, discoverable catalog of your entire data landscape. This system automates data lineage tracking and enforces governance policies across all layers. For ultimate trust and compliance, critical data events are certified on a Private Blockchain, creating an immutable, cryptographically verifiable audit trail to satisfy the most stringent regulatory requirements like GDPR and SOX.
Performance and Flexibility
The architecture is designed for performance and flexibility. A cross-cutting query engine powered by Trino provides high-performance, federated SQL access across all data layers, enabling complex analytics without costly data movement. Apache Flink enables real-time stream processing for immediate insights, feeding directly into the Standardized layer alongside traditional batch ingestion.
The Economic Performance Metrics Layer serves as the optimized source for all downstream applications, from interactive BI Dashboards and predictive Machine Learning Models to real-time Data APIs.
The Interval Data Portal
All derived value—BI intelligence, predictive insights, API outputs, and verifiable provenance from the blockchain—converges in the Interval Data Portal. This enterprise-facing hub provides a single, consolidated, and secure access point for all your enterprise’s data-driven intelligence needs.
Global Reach, Local Control
Sovereign Data, Global Intelligence
Modern enterprises face a fundamental paradox. On one hand, the promise of AI and global analytics demands access to vast, interconnected datasets to drive innovation and maintain a competitive edge. On the other, a growing patchwork of data privacy regulations—such as GDPR, CCPA, HIPAA, and PIPA—mandates strict controls over data residency, requiring that sensitive information remain within specific geographic or jurisdictional boundaries.
Interval addresses this challenge with a federated architecture that brings intelligence to the data, rather than moving data to the intelligence.
Foundation: Private Storage for Compliance
Interval’s architecture is built on the principle that raw, sensitive data should never leave its owner’s control. The foundation of the system is the Harvest platform, a private, single-tenant infrastructure deployed directly within an enterprise’s own cloud environment or on-premises servers.
Private, Dedicated Data Lakehouse: Unlike multi-tenant public cloud solutions, each Interval customer is provisioned with a private, single-tenant Cognitive Data Lakehouse. This isolated environment provides maximum security and control.
Flexible, Compliant Deployment: The Harvest instance can be deployed on-premises within an enterprise’s own data center or within a specific cloud region (e.g., Frankfurt for EU data). This flexibility is critical for meeting strict data residency requirements.
Security by Design: Data is protected with end-to-end encryption, both in transit (TLS 1.3) and at rest (AES-256-GCM).
Search: A Federated Network for Global Queries
Global queries in Interval work by sending the question to the data, not the data to a central system.
A user (or application) sends a question—for example: “What was our global churn rate last quarter by region?”—to the Interval platform.
The Intelligence Agent Framework analyzes the request, decides which regions and Harvest instances are relevant, and generates a distributed query plan.
AI Agents are dispatched to each regional Harvest deployment, where they run the query locally against Standardized and Interval Data Standard data. Raw records never leave the region.
Each region returns only aggregated, privacy-preserving results (e.g., counts, rates, KPIs), which are then combined into a single global answer.
Throughout this process, every significant data operation is anchored to the Interval EVM Blockchain. Each execution is:
Recorded via EAS as a Data Story describing what was computed and from which logical dataset.
Logged via EAS to provide a chronological, immutable audit trail.
This design gives you global analytics with local control: you can run cross-region queries with full observability into how, where, and on what data they were executed—without ever centralizing sensitive raw data.
Exchange: Secure Egress of Derivative Data
The same mechanism that powers global queries also underpins how Interval shares insight externally. While raw data remains local, the platform can package the derived results of these federated computations as shareable, monetizable assets.
Instead of exposing source tables, Interval exposes secure, derivative data products, such as:
AI-Generated Insights & Reports: Aggregated analytics delivered through the secure Interval Portal or APIs.
Private Model Artifacts: Models trained on Standardized and Interval Data Standard assets inside each region, so the models learn from your data without copying raw records out.
Licensed Data Products: Outputs wrapped in ERC-5006 License NFTs, which encode usage terms such as seat limits, duration, geography, or level of detail.
These derivative assets can be exchanged with partners or customers via the Interval Exchange without ever exposing underlying raw data. By processing data locally and sharing only privacy-preserving outputs, Interval enables organizations to participate in the global data economy on their own terms—securely, compliantly, and with full auditability.
Agent Framework Architecture
Enterprise-Grade AI-Driven Orchestration
The Interval Agent Framework is designed to help enterprises harness the power of artificial intelligence without compromising data security or requiring massive technical transformation. It addresses the critical business challenge of unlocking value from siloed data by bringing intelligence directly to the source.
The architecture is built upon a foundation of multi-layered security that spans from the network edge to the application layer:
Edge Security: Cloudflare provides DDoS protection and WAF filtering.
Identity Management: Keycloak handles OAuth flows and RBAC policies in isolation.
Runtime Security: All operations run within a Hardened Private Kubernetes Cluster secured with a mandatory mTLS Service Mesh.
Multi-Agent Intelligence
Business inquiries are processed by a coordinated team of four specialized AI agents:
Data Engineer Agent: Safely connects to existing enterprise systems (e.g., SAP, Oracle), then structures, cleans, and prepares data with privacy-preserving safeguards.
Data Analyst Agent: Organizes the prepared data into meaningful business relationships.
Context Agent: Leverages its understanding of industry, regional regulations, and local business practices to interpret complex business conversations.
BI Agent: Produces executive-level recommendations and actionable insights from the structured data.
Safety and Determinism
At the core of the framework is LangGraph, serving as both the State Orchestration and Execution Engine. It manages complex agentic workflows, maintains state persistence, and coordinates AI-driven decision points.
To ensure accuracy and mitigate risks like data leakage or AI hallucinations, all agent interactions with AI models are passed through Prompt Guardrails and Consensus & Quality Filters. The framework offers complete Model Flexibility, allowing enterprises to use privately-hosted models for maximum security or external providers for specific capabilities.
Last updated